The final Year 3 research project at NUA. Made within 8 weeks in a language and engine I had little prior knowledge in: Monogame & Nez.
Highlights:
C#
A* Pathfinding
Data Generation
Navigation
AI
C#
A* Pathfinding
Data Generation
Navigation
AI
Software/Frameworks:
Monogame & Nez
Visual Studio
Monogame & Nez
Visual Studio
The goal was to create pathfinding where no manual graph modification or recording player input is required.
In order to do this, many jump trajectories are generated with various inputs, which are then used to link ground nodes together using an AABB sweep.
Once the ground nodes are linked together, the resulting graph can be parsed to A*, returning a path for the AI to follow.
A simple state machine navigates an agent to the points on the path and if it needs to jump, the agent's inputs are set to the recorded inputs from when jump trajectories were generated.
No path takes more than a millisecond to find. However, the initial data generation for the graph takes around 2-3 seconds on a low-end computer.
Even then, the benefit of being able to change levels or modify character properties without having to manually fix the pathfinding graph later possibly outweighs the 2-3 second generation time.
